Abstract

In this thesis, we study the integration of data and computational methods in nonlinear system analysis and control design. In the first part of this thesis, we develop a constructive framework for designing a set of candidate functions from data such that desired control-theoretic conditions are satisfied. This new design framework is aimed at offering a general data-driven approach to a number of problems in systems and control, including stability analysis, stabilization, and safety-critical control designs. The second half of the thesis focuses on the specific problem of designing a point-to-point steering, which is often considered in many practical control applications. For this problem, we introduce a highly integrated and data-efficient control approach that directly integrates raw data from output measurements into the control synthesis and tailors the acquisition of data toward the control design. This approach is further extended to safely perform learning in unknown constrained environments where the constrained region might not be pre-specified in advance and can only be identified when the system approaches those areas. Practical aspects of data, such as data acquisition challenges and deficient data quality, are carefully considered throughout the development of all solutions to facilitate attainable physical implementation.

Committee Chair

Shen Zeng

Degree

Doctor of Philosophy (PhD)

Author's Department

Electrical & Systems Engineering

Author's School

McKelvey School of Engineering

Document Type

Dissertation

Date of Award

12-22-2023

Language

English (en)

Download

Included in

Engineering Commons

Share

COinS